Deep learning-based particle pairing in single-frame dual-exposure holography
Applied Optics · 10.07.2026 · 65 (20) · 6800
DOI: 10.1364/ao.597729 ↗Kurz gefasst
Fachartikel in der Zeitschrift Applied Optics, online erschienen am 06.07.2026. Verfasst von Kaihang Zheng und 4 weiteren Autor:innen. Die Arbeit nennt 24 Literaturangaben. Gefördert von: Science Challenge Project, National Natural Science Foundation of China.
Abstract (Originalsprache)
This study performs particle matching on single-frame double-exposure holographic data acquired from shock-induced droplet breakup experiments. Volumetric patches of 20×20×3pixels centered on individual particles are extracted from reconstructed holograms and processed by a 3D Siamese network to compute feature similarities. A directional prior is introduced by statistically aggregating the orientation distribution of neighboring particles, which constrains candidate searches and reduces mismatches in dense particle fields. Under the tested conditions, the method achieves 95.27% pairing accuracy on static datasets and 92.14% on dynamic datasets. For reference, normalized cross-correlation (NCC) under the same conditions yields 86.73% accuracy. The described approach provides a practical tool for holographic particle tracking velocimetry (PTV) in scenarios where multi-frame acquisition is challenging.
Förderung
- Science Challenge Project
- National Natural Science Foundation of China
Zitieren
APA
Zheng, K., Zhao, Y., Cai, Q., Liu, J., & Bao, L. (2026). Deep learning-based particle pairing in single-frame dual-exposure holography. Applied Optics, 65(20), 6800. https://doi.org/10.1364/ao.597729
Vancouver
Zheng K, Zhao Y, Cai Q, Liu J, Bao L. Deep learning-based particle pairing in single-frame dual-exposure holography. Applied Optics. 2026;65(20):6800. doi:10.1364/ao.597729
BibTeX
@article{Zheng2026,
author = {Zheng, Kaihang and Zhao, Yu and Cai, Quancheng and Liu, Jinhong and Bao, Lijun},
title = {Deep learning-based particle pairing in single-frame dual-exposure holography},
journal = {Applied Optics},
year = {2026},
volume = {65},
number = {20},
pages = {6800},
doi = {10.1364/ao.597729}
}
EUVIMED